Spaces:
Build error
Build error
File size: 5,332 Bytes
557ef1f |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 |
from __future__ import print_function
from src.misc.config import cfg, cfg_from_file
from src.dataset import TextDataset
from src.trainer import condGANTrainer as trainer
import time
import random
import pprint
import numpy as np
import torch
import torchvision.transforms as transforms
from pathlib import Path
import streamlit as st
def gen_example(wordtoix, algo, text):
"""generate images from example sentences"""
from nltk.tokenize import RegexpTokenizer
data_dic = {}
captions = []
cap_lens = []
sent = text.replace("\ufffd\ufffd", " ")
tokenizer = RegexpTokenizer(r"\w+")
tokens = tokenizer.tokenize(sent.lower())
rev = []
for t in tokens:
t = t.encode("ascii", "ignore").decode("ascii")
if len(t) > 0 and t in wordtoix:
rev.append(wordtoix[t])
captions.append(rev)
cap_lens.append(len(rev))
max_len = np.max(cap_lens)
sorted_indices = np.argsort(cap_lens)[::-1]
cap_lens = np.asarray(cap_lens)
cap_lens = cap_lens[sorted_indices]
cap_array = np.zeros((len(captions), max_len), dtype="int64")
for i in range(len(captions)):
idx = sorted_indices[i]
cap = captions[idx]
c_len = len(cap)
cap_array[i, :c_len] = cap
name = "output"
key = name[(name.rfind("/") + 1) :]
data_dic[key] = [cap_array, cap_lens, sorted_indices]
algo.gen_example(data_dic)
# streamlit function
def center_element(type, text=None, img_path=None):
"""
Function to center a streamlit element (text, image, etc)
"""
if type == "image":
col1, col2, col3 = st.columns([1, 2, 1])
elif type == "text" or type == "heading":
col1, col2, col3 = st.columns([1, 6, 1])
elif type == "subheading":
col1, col2, col3 = st.columns([1, 2, 1])
elif type == "title":
col1, col2, col3 = st.columns([1, 8, 1])
with col1:
st.write("")
with col2:
if type == "heading":
st.header(text)
elif type == "title":
st.title(text)
elif type == "image":
st.image(img_path)
elif type == "text":
st.write(text)
elif type == "subheading":
st.subheader(text)
# else:
# raise Exception("Unsupported input type")
with col3:
st.write("")
def demo_gan():
cfg_from_file("eval_bird.yml")
# print("Using config:")
# pprint.pprint(cfg)
cfg.CUDA = False
manualSeed = 100
random.seed(manualSeed)
np.random.seed(manualSeed)
torch.manual_seed(manualSeed)
output_dir = "output/"
split_dir = "test"
bshuffle = True
imsize = cfg.TREE.BASE_SIZE * (2 ** (cfg.TREE.BRANCH_NUM - 1))
image_transform = transforms.Compose(
[
transforms.Resize(int(imsize * 76 / 64)),
transforms.RandomCrop(imsize),
transforms.RandomHorizontalFlip(),
]
)
st.cache(func=TextDataset, persist=True,ttl=10000)
dataset = TextDataset(
cfg.DATA_DIR, split_dir, base_size=cfg.TREE.BASE_SIZE, transform=image_transform
)
assert dataset
dataloader = torch.utils.data.DataLoader(
dataset,
batch_size=cfg.TRAIN.BATCH_SIZE,
drop_last=True,
shuffle=bshuffle,
num_workers=int(cfg.WORKERS),
)
# Define models and go to train/evaluate
st.cache(
func=trainer, persist=True, suppress_st_warning=True,ttl=10000
)
algo = trainer(output_dir, dataloader, dataset.n_words, dataset.ixtoword)
st.title("Text To Image Generator ")
st.subheader("Enter the description of the bird in the text box you like !!!")
st.write(
"**Example**: A yellow bird with red crown, black short beak and long tail"
)
st.markdown("**PS**: The synthesized birds might not even exist on earth ")
st.markdown("#")
user_input = st.text_input("Write the bird description below")
st.markdown("---")
if user_input:
start_t = time.time()
# generate images for customized captions
gen_example(dataset.wordtoix, algo, text=user_input)
end_t = time.time()
print("Total time for training:", end_t - start_t)
st.write(f"**Your input**: {user_input}")
center_element(type="subheading", text="AttnGAN synthesized bird")
st.text("")
center_element(
type="image", img_path="models/bird_AttnGAN2/output/0_s_0_g2.png"
)
center_element(type="subheading", text="The attention given for each word")
st.image("models/bird_AttnGAN2/output/0_s_0_a1.png")
st.markdown("---")
with st.expander("Click to see the first stage images"):
st.write("First stage image")
st.image("models/bird_AttnGAN2/output/0_s_0_g1.png")
st.write("First stage attention")
st.image("models/bird_AttnGAN2/output/0_s_0_a0.png")
def attngan_explained():
# center_element(type="heading", text="AttnGAN: Fine-Grained Text To Image Generation with Attentional Generative Adverserial Networks")
st.header(
"**AttnGAN**: Fine-Grained Text To Image Generation with Attentional Generative Adverserial Networks"
)
from attngan_explanation import attngan_explanation
attngan_explanation()
|